Download Open issues in routing techniques in ad hoc wireless sensor networks

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Policies promoting wireless broadband in the United States wikipedia , lookup

Distributed firewall wikipedia , lookup

Network tap wikipedia , lookup

IEEE 802.1aq wikipedia , lookup

Zero-configuration networking wikipedia , lookup

IEEE 1355 wikipedia , lookup

Wireless security wikipedia , lookup

Computer network wikipedia , lookup

Peering wikipedia , lookup

Piggybacking (Internet access) wikipedia , lookup

Cracking of wireless networks wikipedia , lookup

Recursive InterNetwork Architecture (RINA) wikipedia , lookup

Peer-to-peer wikipedia , lookup

Airborne Networking wikipedia , lookup

Routing wikipedia , lookup

Routing in delay-tolerant networking wikipedia , lookup

Transcript
Open issues in routing techniques in ad hoc wireless
sensor networks
Nitin Nagar and Edoardo S. Biagioni
{nitin,esb}@hawaii.edu
Department of Information & computer sciences
University of Hawaii at Manoa, Honolulu, HI
Abstract:
Ad hoc sensor networks are ad hoc networks that
are characterized by decentralized structure and
ad hoc deployment. Sensor networks have all the
basic features of ad hoc networks but to different
degrees – for example, much lower mobility and
much more stringent energy requirements. We
analyze the current state of research and evaluate
open issues in development of routing techniques
in wireless sensor networks.
Keywords: General Sensor Network,
MANETs, PODS
Introduction
Consider a wireless network made up of units
that perform both measurements and
communications. These units are totally
independent and are capable of recording
data from sensors. The mobility of these
units is very low but the data forwarding
strategy is robust enough to be fault tolerant
and to allow occasional mobility among
units.
For example, we consider the sensor network
being developed for the PODS project at UH
Manoa [9]. One of the major objectives of
the project is to implement a sensor network
to study endangered plants such as “Silene
Hawaiiensis”, in order to determine what is
essential for the plant’s survival in its native
habitat. The challenge is to implement an ad
hoc network comprised of hundreds of small
sensors or pods, which monitor wind, rain,
temperature, light and moisture, and which
are used for determining spatial or temporal
patterns in the environment of the plant being
studied.
Such a real life sensor network is comprised
of hundreds of sensors. The nodes are battery
powered, so the first networking challenge is
getting data back with minimal energy
expenditure, by choosing energy-efficient
paths and by minimizing the routing
overhead. The second challenge is to
maintain connectivity in case some pods are
moved to a different location or fail to
participate due to lack of power, though
overall mobility is likely to be more limited
than in a network of laptops. The third
challenge is that sensor networks can be
expected to grow to many thousands of
nodes, so any algorithms used in these
networks must be scalable. Finally, these
networks should use multiple paths whenever
possible, both for redundancy and to
distribute the energy expenditure of
forwarding packets.
These requirements
distinguish ad-hoc wireless sensor networks
from mobile ad-hoc networks (MANETs).
Table 2 compares MANETs with ad hoc
sensor networks.
A sensor network such as the PODS network
also differs from many of the wireless sensor
networks considered in the literature.
Though some data can be combined and
summarized, other data, for example camera
images, must be delivered unchanged to a
base station. The PODS network is designed
to have multiple base stations if possible. In
addition, communication is not limited to
sending data to base stations: interaction
between individual sensor nodes may be
needed to allow distributed computation
among nodes in close geographic proximity,
to support occasional communication from
the base stations to the individual nodes, and
for a variety of reasons including faulttolerance. These requirements mean that a
PODS-style network needs to be able to
support any-to-any communication, though
the common mode of communication is from
nodes to one or more base stations.
There has been a lot of research in wireless
routing protocols. Existing protocols provide
different tradeoffs among the following
desirable characteristics: fault tolerance,
distributed
computation,
robustness,
scalability, and reliability. Wireless protocols
proposed so far for wireless sensor networks
are very limited, generally focusing on
communication to a single base station or on
merging sensor data. While these protocols
are suitable for their intended purposes, in
this paper we explore the use of protocols
developed for MANETs to provide more
general communication among nodes in a
sensor network.
Low Energy Consumption
Protocols for MANETs are designed for
communication among laptops. Even though
laptops are battery-powered, their power
budget far exceeds that of a node in a
wireless sensor network. Such nodes are
often deployed in remote locations. Whether
powered by batteries, solar energy, or some
other method, reducing energy consumption
lessens the weight or extends the lifetime of
the package and makes the sensor easier to
conceal. Each node in a wireless sensor
network only needs to record, transmit, and
forward data, unlike a laptop which might
have to perform much more complex tasks.
As a result, the computational engine in a
sensor node consumes significantly less
energy than a laptop, and communications
must likewise use less energy.
Many routing techniques have been proposed
for both MANETs and wireless sensor
networks
[3,11,7,12,18,21,22,26,27,28],
including protocols that focus on “minimum
energy” routing [21,23,24,25,27,29]. For
example, [23] notes that a route with more,
shorter hops often requires less total energy
than a route using fewer, but longer, hops.
Other papers focus on developing
generalized power aware/energy aware
routing schemes, designing power aware cost
metrics [23], using transmit power
adjustment to control the network topology
[19], or using the location information to
minimize the power relay route, thus
minimizing the total energy consumption
[24]. However, none of these studies focus
on practical issues such as the overhead of
computing such minimum-energy routes.
Doshi et al., [21] list the following reasons
why minimum energy routing is hard to
implement in practice. Minimum energy
routing introduces an overhead cost, the
additional routing information is not free,
current protocols fail to provide sufficient
information for making power level
decisions, lower power routes leave less
margin for channel fluctuations or
measurement errors, minimum energy routes
are difficult to discover, and minimum
energy routes are difficult to maintain.
Because of these issues, it is not currently
clear that such “minimum energy” routing is
in practice any better than other methods
which have lower theoretical efficiency but
provide other practical advantages. Because
of these limitations, we consider a variety of
protocols, not only those which claim to use
“minimal” energy.
The SensIT program at DARPA notes that
many MANET protocols focus on fast
topology changes, and that a focus on poweraware metrics, location information, and the
energy usage of each node can lead to more
power-aware routing. Location information
is used by some MANET protocols [2, 5, 20,
25, 28, 29] both to improve scalability and, in
some cases, also to minimize energy
consumption [14, 27, 29]. Some authors [29]
point out that an optimal geographic route
may provide power savings and network
lifetime extension compared to a similar
route that does not use location information.
However, at least for the GEAR protocol this
has only been tested under a very limited
number of relatively favorable network
configurations.
LEACH [27] proposes a clustering based
protocol that utilizes randomized rotation of
local cluster heads to evenly distribute energy
load among the sensors in the network. In
LEACH each local cluster head performs
“local data fusion” to compress the
information. It is a single path routing
technique whose scalability is provided by its
hierarchical nature. However, some of their
assumptions may not be true when compared
to general sensor networks like PODS.
LEACH requires a fixed base station to
which data needs to be routed. Leach only
supports sensors which do not move and send
data at fixed rate, with symmetric radio
channels, and adjustable transmit power.
Leach also assumes that the cluster heads can
talk directly to the gateway node. LEACH
may be a good solution for a smaller subset
of the problem but for a general sensor
network like PODS, it does not address the
situation when more than one fixed base
station is present, where sensors are not
static, and where a node can communicate to
another node in an arbitrary fashion.
GEAR [29] presented an alternative to this
by incorporating the technique of data
diffusion and using geographic computations
to find low energy paths. They propose that
if the destination is quite far from the packet
then the path found by geographic routing
may be nearly as energy efficient as an
optimal route. Some of their techniques that
are not useful for the more general sensor
networks we are interested in, include the use
of data diffusion, which is only useful to
deliver the data to a single or a few base
stations,
and
does
not
support
communication between arbitrary nodes. The
paper by Doshi et al. [21] showed that a path
which is primarily discovered using location
information may not be the most energy
efficient path.
Low mobility
Sensor networks differ from MANETs in a
very important way and that is in mobility. A
MANET is a more general case where the
participating laptops can either be stationary
or move randomly with a random speed. As
nodes within a MANET move, they move
out of range of their neighbors and hence are
no longer able to communicate with the old
neighboring nodes and come within range of
new nodes. Hence the mobility introduces the
problem of fault tolerance. An ideal routing
protocol for MANET should be able to
deliver data packets from source to
destination even when some of the
intermediate nodes move away from their
neighbors range. This complicates the design
of the routing protocol as this introduces
additional routing overhead. In previous
work [17], one of the authors related the
speed of the movement of the nodes to the
packet delivery ratio and routing overhead.
The packet delivery ratio worsens as speed is
increased for DSR [8], whereas AODV [6]
does not degrade as rapidly when mobility
increases. Nodes in a sensor network most of
the time are static and with an occasional
breaking of a link as the node runs out of its
energy or is relocated. Sensor networks need
the ability to re-configure automatically in
case links disappear or new nodes appear.
Protocols such as GEAR and LEACH
assume that the nodes in a sensor network are
static where as in PODS at least some of the
nodes (e.g. a hand-held base station) may be
mobile.
Self-configuring nature
Ad hoc wireless sensor networks are selfconfiguring in nature. This can be considered
an added feature to the existing ad hoc nature
of the network. The network is adaptable to
the changing requirements and is able to
diagnose when a link / sensor node goes
down and when it comes up. There are two
main schemes to design a wireless sensor
network, the address centric scheme and the
data centric scheme. The address centric
scheme has been used by various routing
protocols such as LAR, GSPR, and DREAM
etc. In this scheme we assign IP addresses to
each sensor node, simplifying the process of
routing. This concept is similar to that of
normal wired networks. A unique IP address
will help the source sensor node to know the
sensor node to which data must be routed.
However in [5] a new concept of data centric
model is presented which is not address
oriented. The mechanism and goal of selfconfiguration in these networks is different
from those of the address centric scheme.
Multipath desirable
Macker and Corson [13] listed qualitative
and quantitative independent metrics for
judging the performance of mobile ad hoc
network routing protocols. One of these
qualitative metrics was path strategy. There
are a number of different path strategies. One
that is very common is shortest path [2, 8, 21,
and 28] where one copy of the message is in
the network at any time. At the other
extreme is the flooding based approach [10,
20] where the message is flooded through the
whole network area. A good example of this
approach is the Multi-path On-demand
Routing (MOR) Protocol [10] which is a ondemand, load balancing routing protocol
designed for the PODS project at the
University of Hawaii at Manoa. MOR may
require as little as one network flood to
establish necessary routes and its energy
efficient and robust in low mobility and low
energy
networks
such
as
PODS.
Broadcasting usually solves the routing in
highly mobile conditions but considering our
requirement for a general sensor network for
PODS this is undesirable. The compromise
between these two approaches is a multipath
strategy, where data packets are routed
through a few distinct paths and successive
packets follow different paths whenever
possible. This not only provides robustness to
the network using multiple paths but also
helps in distribution of the energy
requirement of the network evenly across the
network. In [1] A.Nasipuri et al., prove that
the use of multiple paths in DSR can keep
correct end to end connections, but they did
not study the performance improvement on
network load balancing. M.R Perlman et al.,
demonstrate [16] that multipath routing can
balance loads. They propose a diversity
injection method to find more node-disjoint
paths compared to DSR. However, their
work is based on multiple channel networks,
which are contention free but may not be
available in some applications. [14] applies
the multipath strategy to DSR’s source
routing technique and achieves some
scalability
under
mobile
conditions.
However the energy distribution component
of the multipath strategy has not been
adequately explored in the paper.
Scalability
An ideal routing protocol for a MANET
should be scalable. This means that as the
size of the network increases or the number
of nodes increases the routing protocol
should be able to adapt to the changes and
provide consistent performance based on the
parameters that we have discussed earlier. [2]
describes three methods, which have been
used by researchers to provide scalability to a
routing protocol for MANETs. The first
method uses hierarchy to provide scalability.
The second way to provide scalability is
caching. The third way to provide scalability
is using geographic information. Using
hierarchy to provide scalability is the most
widely deployed approach to scale routing as
the number of destinations increases. Two
main strategies used to combine nodes
location and hierarchical network structures
are the Zone Based Routing and the
Dominating Set Routing. Online poweraware routing routing [28] schemes are
example of Zone Based Routing and GRID is
an example of dominating set routing.
Caching is becoming a widely deployed
technique for scaling ad hoc routing
protocols in MANET [6, 8]. Caching reduces
the routing protocols message load in two
ways: It avoids pushing topological
information where the forwarding load does
not require it (like ideal routers) and it often
reduces the number of hops between the
router that has topological information and
the router that requires it. However, Doshi et
al.,[21]
demonstrated
with
their
implementation of energy aware DSR
protocol using old routes from the cache does
not necessarily mean that a low energy route
is selected every time.
The last and most frequently used technique
to provide scalability to ad hoc routing
protocols is to use the geographic location
information. This technique assumes that all
wireless nodes know their positions and links
are bi-directional. This approach has been
adapted in GPSR, GEAR and gradient
routing.
For a general sensor network a combination
of the above-mentioned strategies would be
sufficient to provide scalability, as mobility
is limited in these networks.
Conclusion
In this paper we have evaluated the necessary
features of the routing protocols for general
wireless sensor network like PODS [9].
Current research into routing protocols for
MANETs and ad hoc sensor networks tend to
make many tradeoffs in various features and
are generally tested in a much regulated
environment. As seen from the paper that the
needs and requirements of routing protocols
for general ad hoc sensor networks is very
unique compared to routing protocols for
MANETs and other sensor networks. Hence,
there is a need for further research into this
new field as it poses some of its unique
challenges and we would be continuing our
research in this area in future.
References
[1] A.Nasipuri and S.R Das, “On demand
Multipath Routing for Mobile Ad Hoc
Networks”, Proceedings of the 8th
International Conference on Computer
Communications and Networks (IC3N),
Boston, 1999
[2] B.Karp and HT.Kung, “GPSR: Greedy
peimeter stateless routing for wireless
networks”, Proc. MOBICOM, August 2000,
pp 243-254.
[3] B.Krishnamachari et al., “Modeling DataCentric Routing in Wireless sensor networks”
[4] C.E.Jones et al.,”A Survey of energy
efficient network protocols for wireless
networks”
[5] Chalermek Intanagonwiwat, R.Govindan
and D.Estrin, “Directed Diffusion: A scalable
and robust communication paradigm for
sensor
networks”,
in
proc.
ACM
MOBICOMM, Boston, MA, 2000
[6] C.Perkins, E.Royer, “Ad hoc on demand
distance vector routing”, Proc. 2nd IEEE
Workshop on Mobile Computer System and
Applications, February 1999, pp. 90-100.
[7] D.Estrin, R.Govindan, J.Heidemann,
S.Kumar, “Next century challenges: Scalable
co-ordination
in
sensor
networks”,
Proceedings of MOBICOM, 1999, Seattle,
263-270.
[8] D.Johnson , D.Maltz, “Dynamic source
routing in ad hoc wireless networks”, Mobile
computing, Chapter 5, pp. 153-181, Luwer
Academic Publishers, 1996.
[9] Edoardo Biagioni, Kent Bridges and
B.J.S Chee, “PODS: A remote ecological
Micro
sensor
network
“,
http://www.botany.hawaii.edu/pods/overview
.htm
[10]Edoardo Biagioni, Kent Bridges, “The
Application of Remote Sensor Technology to
Assist the Recovery of Rare and Endangered
Species”, Accepted for publication in a
special issue on Distributed Sensor Networks
for the International Journal of High
Performance Computing Applications, Vol.
16, N. 3 (August 2002).
[20] S.Basagni, I.Chtamtac, V.R.Syrotiuk,
B.A Woodward, “A distance routing effect
algorithm
for
mobility
(DREAM)”,
Proceedings of MOBICOM, 1998, 76-84.
[11] I.D.Aron., “On the scalability of ondemand routing protocols for mobile ad hoc
networks: an analytical study”
[21] S.Doshi, T.X.Brown, “Minimum energy
routing schemes for a wireless ad hoc
network”, IEEE INFOCOM 2002.
[12] I.Stojmenovic and Xu Lin, “Poweraware localized routing in wireless
networks”, IEEE Int. Parallel and distributed
processing symp., Cancun, Mexico, May 1-5,
2000, pp. 371-376.
[22]
S.Giordano,
I.Stojmenovic
et
al.,”Position based Routing: Algorithms for
Ad Hoc Networks: ATaxonomy”
[13] J.P. Macker and M.S. Corson, “Mobile
ad-hoc networking and the IETF”, Mobile
computing and communications review, 2, 1,
1998, 9-14.
[14] K.Wu and J.Harms, “On-demand
multipath routing for mobile ad-hoc
networks”, Computer science department,
University of Alberta, AB, Canada.
[15] M.Mauve, J.Widmer, “A Survey on
Position-Based Routing in Mobile Ad Hoc
Networks”
[16] M.R Perlman, Z.J.Hass, P.Sholander and
S.S Tabrizi, “On the impact of Alternate Path
Routing for load balancing in Mobile Ad Hoc
Networks”, Proceedings of IEEE/ACM
MobiHoc 2000, Boston, 2000
[17] Nitin Nagar and A.Kongmunvattana,
"Analysis of Routing Protocol Performance
on Multi-Hop Wireless Ad Hoc Networks",
The 2001 International Conference on
Parallel
and
Distributed
Processing
Techniques and Applications (PDPTA'2001),
Las Vegas, June 25-28, 2001
[18] Robert D.Poor, “Gradient Routing in Ad
Hoc Networks”
[19]
R.Ramanathan,
R.Rosales-hain,
“Topology control of multi-hop wireless
networks using transmit power adjustment”,
IEEE INFOCOM 2000, pp. 404-413.
[23] S.Singh, M., and C.S.Raghavendra,
“Power aware routing in mobile ad-hoc
networks”, Proceedings of MOBICOM,
1998, pp. 181-190.
[24] V.Ruduplu, T.Meng, “Minimum energy
mobile wireless networks”, IEEE JSAC, v.17,
n.8, August 1999, pp. 13333-44.
[25] W.H. Liao, Y.C. Tseng. J.P. Sheu,
“GRID: A fully location-aware routing
protocols for mobile ad-hoc networks”, Proc.
IEEE HICSS, January 2000.
[26] W.R.Heinzelman, A.Chandrakasan and
H.Balakrishnan, “Energy efficient routing
protocols
for
wireless
microsensor
networks”, Proc. HICSS, Hawaii, January
2000.
[27]
W.,
Rabiner,
A.Heinzelman,
A.Chandrakasan
and
H.Balakrishnan,
“Energy-efficient communication protocol
for wireless microsensor networks”, in
proceedings of the 33rd HICSS, 2000
[28] Y.B.Ko and N.H Vaidhya, “Locationaided routing (LAR) in mobile ad hoc
networks”, Proc. MOBICOM, 1998, pp. 6675.
[29] Yan Yu, R.Govindan, D.Estrin,
“Geographical and Energy Aware
Routing: a recursive data dissemination
protocol for wireless sensor networks”
Table 1: Comparison of three different energy efficient routing protocols for ad hoc
sensor networks
Method
Distributed
Metrics
Scalability
New
DSR
[4]
LEACH
[19]
No
Shortest
path
Yes :
Location
aware:
clustering
Yes:
Location
aware
GEAR
[28]
Multiple
base
stations
Yes
Mobility
Fault
Tolerance
Robustness
Caching
Comm.
Without
merging
Yes
Sufficient
Single
path
Yes:
Limited
Location,
Hierarchical
No
No
No
Single
path
No
Shortest
Path,
Location
Geographic
No
Yes
Not
tested
Single
path
No
Table 2: Comparison of features of MANETs and Ad Hoc Sensor Networks.
Features
Decentralized control
Bandwidth deficient
Energy deficient
Mobility
Traffic
Data rate
Flow of data
MANET
Yes
Yes
Yes. But this is of secondary
importance as battery packs can
be replaced
Varies (slow to fast)
Multimedia rich
High
Bi-directional
Redundancy in data
Main Goal
No
To optimize QoS and high
bandwidth efficiency
Fault tolerance
Needed as mobility increases
Basic features of routing protocol
Loop free, energy and bandwidth
efficient, secure, provides QoS,
fault tolerant and reactive instead
of proactive, and distributed in
nature
Ad Hoc Sensor Networks
Yes
Sometimes
Yes, it is of primary importance
Limited
Statistical and Multimedia
Low [1-1000 Kbps]
Mostly uni-directional [sensor to
sink]
Sometimes
Prolonging the life of the network
through aggressive energy
management, to prevent
connectivity degradation.
Needed only if nodes exhaust
available energy or are moved
Most of the same features as for
MANETs, but with less emphasis
on mobility and more emphasis
on energy efficiency, scalability,
and multipath connectivity.